• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Development of an integrated method between deep learning and physical models in lowland drainage management with insufficient observed data

Research Project

Project/Area Number 21K05838
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Review Section Basic Section 41030:Rural environmental engineering and planning-related
Research InstitutionNational Agriculture and Food Research Organization

Principal Investigator

Kimura Nobuaki  国立研究開発法人農業・食品産業技術総合研究機構, 農村工学研究部門, 上級研究員 (40706842)

Project Period (FY) 2021-04-01 – 2024-03-31
Project Status Completed (Fiscal Year 2023)
Budget Amount *help
¥2,990,000 (Direct Cost: ¥2,300,000、Indirect Cost: ¥690,000)
Fiscal Year 2023: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Fiscal Year 2022: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2021: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Keywords少ないデータ量 / 深層学習 / 水位予測手法 / 物理モデル / 転移学習 / 少ない量のデータ / 排水管理 / データ補完技術
Outline of Research at the Start

効率、かつ迅速な低平地の排水管理を行なうために、深層学習モデルを導入した水位予測手法が必要であるものの、深層学習モデルの実行に不可欠な良質、かつ大量の観測データの収集に時間と労力を奪われることが課題である。これを解決するために、物理モデルを導入し、観測値の代替となる擬似データを生成する。さらに、あるデータのパターンを別のデータに転移する「転移学習」を導入することで、疑似データを観測値に近似できる。このように、物理モデルに起因する誤差を含む疑似データをより現実的なデータに修正し、データ補充を行うことで、観測データが不十分な場合にも物理モデルと深層学習モデルによって高精度な水位予測を可能にする。

Outline of Final Research Achievements

This study aimed to create a robust AI model that predicts water levels for0 severe flood events in rivers and agricultural water facilities (e.g., drainage pumping stations) using deep learning. For a small number of data samples, such as the number of large-flood occurrences, the AI models have poor performance in accurate prediction due to data-driven models that require large amounts of data. To improve this problem, first, we artificially generated a large amount of virtual data comparable to large floods using a physical model (e.g., a runoff analysis model), and used the data as training data for the AI model to construct a pre-learned model. Next, to incorporate with the observed data features, we established a highly accurate prediction method by retraining part of the pre-trained model (i.e., transfer learning) using few observed data samples.

Academic Significance and Societal Importance of the Research Achievements

本研究の学術的意義について、一般に、深層学習モデルは、データサンプル数が少ない場合には、予測精度が劣るものの、その欠点を補うために物理モデルからの疑似データを割増し、さらに、転移学習で疑似データの特徴をサンプル数が少ない対象に転移することで、予測精度の向上が可能な手法(物理ガイド深層学習モデル)を開発した。
社会的意義について、現地の観測データのサンプルが少ない場合でも、物理モデルで疑似生成された大量のデータを併用して学習する、物理ガイド深層学習モデルは、現地への適用を通して、実用的に有用であることを明らかにし、さらに、データ保有に関して同様な条件の他の地区へ普及させる可能性が見出せた。

Report

(4 results)
  • 2023 Annual Research Report   Final Research Report ( PDF )
  • 2022 Research-status Report
  • 2021 Research-status Report
  • Research Products

    (14 results)

All 2024 2023 2022 2021 Other

All Journal Article (7 results) (of which Peer Reviewed: 7 results,  Open Access: 2 results) Presentation (6 results) (of which Int'l Joint Research: 2 results,  Invited: 1 results) Remarks (1 results)

  • [Journal Article] WATER LEVEL PREDICTIONS THAT VISUALIZE UNCERTAINTY USING LSTM COUPLED WITH ALTERNATIVE METHODS IN BAYESIAN INFERENCE AT A RESERVOIR NEAR A DRAINAGE PUMPING STATION2024

    • Author(s)
      木村延明,皆川裕樹,福重雄大,吉永育生,馬場大地
    • Journal Title

      Japanese Journal of JSCE

      Volume: 80 Issue: 22 Pages: n/a

    • DOI

      10.2208/jscejj.23-22011

    • ISSN
      2436-6021
    • Related Report
      2023 Annual Research Report
    • Peer Reviewed
  • [Journal Article] CREATION AND VERIFICATION OF A PRETRAINED MODEL FOR RIVER FLOOD PREDICTIONS2024

    • Author(s)
      木村延明,皆川裕樹,福重雄大,馬場大地
    • Journal Title

      Japanese Journal of JSCE

      Volume: 80 Issue: 16 Pages: n/a

    • DOI

      10.2208/jscejj.23-16147

    • ISSN
      2436-6021
    • Related Report
      2023 Annual Research Report
    • Peer Reviewed
  • [Journal Article] A sophisticated model for riverine-flood predictions using convolutional LSTM and transfer learning2023

    • Author(s)
      木村 延明、皆川 裕樹、福重 雄大、吉永 育生、馬場 大地
    • Journal Title

      Artificial Intelligence and Data Science

      Volume: 4 Issue: 3 Pages: 361-368

    • DOI

      10.11532/jsceiii.4.3_361

    • ISSN
      2435-9262
    • Related Report
      2023 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Examining practical applications of a neural network model coupled with a physical model and transfer learning for predicting an unprecedented flood at a lowland drainage pumping station2023

    • Author(s)
      Kimura Nobuaki、Minakawa Hiroki、Kimura Masaomi、Fukushige Yudai、Baba Daichi
    • Journal Title

      Paddy and Water Environment

      Volume: 21(4) Issue: 4 Pages: 509-521

    • DOI

      10.1007/s10333-023-00944-8

    • Related Report
      2023 Annual Research Report
    • Peer Reviewed
  • [Journal Article] GENERATING AN AI PRETRAINED MODEL FOR FLOOD PREDICTIONS USING TRANSFER LEARNING2023

    • Author(s)
      木村 延明、皆川 裕樹、福重 雄大、馬場 大地
    • Journal Title

      Advances in River Engineering

      Volume: 29 Issue: 0 Pages: 79-84

    • DOI

      10.11532/river.29.0_79

    • ISSN
      2436-6714
    • Related Report
      2023 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] EVALUATION OF A WATER-LEVEL PREDICTION MODEL THAT EMPLOYS SUPPORT VECTOR REGRESSION IN WATER & DRAINAGE MANAGEMENT2022

    • Author(s)
      木村延明, 皆川裕樹, 福重雄大, 馬場大地
    • Journal Title

      Journal of Japan Society of Civil Engineers, Ser. B1 (Hydraulic Engineering)

      Volume: 78 Issue: 2 Pages: I_139-I_144

    • DOI

      10.2208/jscejhe.78.2_I_139

    • ISSN
      2185-467X
    • Related Report
      2022 Research-status Report
    • Peer Reviewed
  • [Journal Article] WATER LEVEL PREDICTIONS IN A DRAINAGE PUMPING STATION USING A DEEP LEARNING MODEL, COUPLED WITH A PHYSICAL MODEL AND A TRANSFER LEARNING APPROACH2021

    • Author(s)
      木村延明, 皆川裕樹, 福重雄大, 木村匡臣, 馬場大地
    • Journal Title

      Journal of Japan Society of Civil Engineers, Ser. B1 (Hydraulic Engineering)

      Volume: 77 Issue: 2 Pages: I_319-I_324

    • DOI

      10.2208/jscejhe.77.2_I_319

    • NAID

      130008160116

    • ISSN
      2185-467X
    • Related Report
      2021 Research-status Report
    • Peer Reviewed
  • [Presentation] Generating an AI-based pretraining flood database (AI-flood DB) using a transfer learning approach2023

    • Author(s)
      Nobuaki KIMURA
    • Organizer
      Asia Oceania Geosciences Society (AOGS) 2023, Singapore
    • Related Report
      2023 Annual Research Report
    • Int'l Joint Research
  • [Presentation] 転移学習を用いた河川洪水イベントの事前学習モデルの構築と検証2023

    • Author(s)
      木村延明
    • Organizer
      農業農村工学会(2023年度大会講演会)
    • Related Report
      2023 Annual Research Report
  • [Presentation] 排水機場遊水地のAI水位予測の高度化2023

    • Author(s)
      木村延明
    • Organizer
      山口大学グローカル環境・防災学研究会、第3回「建設分野におけるAI活用の最前線」
    • Related Report
      2023 Annual Research Report
    • Invited
  • [Presentation] ベイズ推定の近似手法を用いた不確実性を考慮したニューラルネットワーク予測モデルの基礎的検討2023

    • Author(s)
      木村延明
    • Organizer
      農業農村工学会 応用水理研究部会 (R5年度講演会)
    • Related Report
      2023 Annual Research Report
  • [Presentation] An inexperienced flood prediction at a drainage pumping station in lowland using an advanced neural network model, coupled with a physical model and transfer learning.2022

    • Author(s)
      Nobuaki KIMURA & Hiroki MINAKAWA
    • Organizer
      PAWEES International Conference 2022
    • Related Report
      2022 Research-status Report
    • Int'l Joint Research
  • [Presentation] 層学習に物理モデルを結合させた排水機場水位予測手法の構築2021

    • Author(s)
      木村延明,皆川裕樹,福重雄大,馬場大地
    • Organizer
      第70回農業農村工学会大会講演会
    • Related Report
      2021 Research-status Report
  • [Remarks] Researchmap

    • URL

      https://researchmap.jp/nkimura3

    • Related Report
      2023 Annual Research Report

URL: 

Published: 2021-04-28   Modified: 2025-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi